Agency

In this section, I explore arguments based on views about agency against both the thesis that institutional actional events qualify as actions, and the thesis that institutional entities can have moral obligations and bear moral responsibility.

In the most familiar cases, an action is or is constituted by an event that is under the inten­tional control of a person and, in virtue of this, the person has the property of doing A, for some action-type A. For example, the event of my writing this sentence is under my intentional con­trol and, in virtue of this, I have the property of writing this sentence, where writing is a type of action. For my purposes, in generalizing from the familiar cases, I need to leave it open whether every action must be constituted by an event that is under the control of a person. So I will say that an action is or is constituted by an event that is under the intentional control of some entity, and, in virtue of this, the entity has the property of doing A, for some action-type A.^13,14

Given even this minimal characterization, there are two ways that one might contest the thesis that institutional actional events are actions. First, one might deny that institutional entities have intentions, and second, one might deny that institutional entities can control events even if they have intentions.

On the above characterization, in order to act, an entity must exercise intentional control over some event, and, I assume, this requires having an intention. One worry is that intentions are states of mind. Given this, entities that lack minds, such as collective entities, also lack intentions.

I take it to be common ground that collective entities do not have minds and are not capable of having conscious phenomenal states of any kind. I am willing to concede that it follows from this that, if collective entities have intentions, their intentions do not have all the properties that the intentions of persons possess. The important point, however, is whether collective entities can be in states that have the key action-relevant properties that are possessed by intentions. Such states would make institutional entities capable of exercising intentional control over an event. And for this, the key factor arguably is whether institutional entities are capable of having plans such that having the plan has an effect on events, or makes a difference to what happens, in the characteristic way.¹⁶ What is this characteristic way?

As Donald Davidson (2001) taught us, a person’s intention can have an effect on events, or make a difference to what happens, even if the person lacks intentional control over these events. For example, a novice skier might form the intention to ski down Headwall Face, but this intention might make her so nervous that she loses control and falls. Her intention made a difference, but not in the way characteristic of intentions. Even if the intention led to her skiing down Headwall Face in accord with her plan, it might not have done this in the characteristic way. Perhaps her intention made her so nervous that she slipped and was forced to ski before she intended to do so. She skied as she intended to, but not when and how she intended. Her skiing was caused by her intention, but not in the way that is characteristic in intentional action.

Unfortunately, it has proven exceedingly difficult to say what this characteristic way is without using the concept of an intention. Fortunately, we can set this aside. Our question is whether an institutional entity can have a plan that has a characteristic kind of effect on actional events involving that entity.

Characteristically, a person with a plan can refer to her plan in deliberation and reasoning, and she can draw on her plan in setting priorities and making decisions about how to go about achieving the goals set by her plan. Further, a person will standardly have non-inferential know­ledge of her plans. Her intentions or plans are standardly “non-inferentially accessible.”¹⁷ Indeed, it is arguable that an agent must be capable of having non-inferentially accessible intentions and beliefs. Suppose that a person intends to ski down Headwall Face. The function of her intention is to bring it about that she skis down Headwall Face. It is to do this by being combined with her beliefs about the circumstances such that, if she comes to believe that the circumstances are right, her intention leads her to intend to ski right then, and, as a result, to start skiing. The ability to bring beliefs and intentions together in deliberation in the way illustrated by the example, and in more complex ways as well, is needed in order for intentions to play their characteristic functional role. This arguably requires an agent’s intentions to be non-inferentially accessible, for otherwise, an agent would not be able to deliberate without making inferences to figure out what she intends. The premises of some such inferences would have to be immedi­ately accessible, on pain of regress (see Ludwig 2007: 424—426).

The non-inferential accessibility of an intention is not the same, however, as its being con­scious or potentially conscious. There is not a recognizable way that having an intention or a plan feels to the person with the intention or the plan, not even when one brings one’s inten­tion “to mind” in order to think about how to carry it out.

I believe, furthermore, that institutional entities are capable of having non-inferentially accessible intentional states. The U.S. knew it was aiming to get Iraq out of Kuwait, and it had this knowledge in an immediately usable way, such that it could monitor what the armed forces were doing to ensure that they were acting efficiently to achieve its goals. If so, it had non-inferentially accessible intentions. In my view, then, institutional entities are capable of having intentions and beliefs to which they have the kind of access needed for reasoning and deliberation.

My reasoning in this section assumes, but also supports, a kind of “functionalism” about intentional states. I think functionalism gives the most plausible account of the nature of such states.¹⁸ As I suggested, the key question for our purposes is whether institutional entities are capable of being in states — having intentions or plans — that make them capable of exercising control of a characteristic kind over actional events involving them, where their being in such states can have an effect on events or make a difference to what happens, in the characteristic way. This is a functional characterization of a kind of state. Even if the intentions ofpersons have non-functional properties that the intentions of institutional entities do not have, the issue is control of a characteristic kind over actional events, and this is a functional matter.¹⁹

Of course, I have not said anything about what the intentions of institutional entities consist in — or about what grounds them.

One might object to my minimal characterization of agency on the grounds that agents, at least in paradigm cases, are responsive to reasons, and they decide what to do based on their understanding of the reasons bearing on their choice. And moral agents are responsive to moral reasons (Wolf 1990; Fischer and Ravizza 1999; Hindriks 2018). This is the kind of control char­acteristic of agency. I agree with this, so I need to address the issue of whether, and if so how, institutional entities respond to reasons, including moral reasons.

There is no puzzle here. A case study would help to make this clear.²⁰ We can imagine the members of the British cabinet, during World War II, deciding not to permit the Royal Air Force to bomb a city for the reason that it is morally wrong to target civilians. Their decision would constitute the cabinet’s deciding against doing something for moral reasons, and presum­ably it would also constitute Britain deciding this.

One might now ask what moral reason the people who serve in relevant roles in an insti­tutional entity would have to pay attention to the moral obligations of the collective. I have already addressed this issue. On the collectivist approach to understanding the moral status of institutional entities, the relevant people have a transitional duty to do their part in ensuring that relevant institutional entities act morally. In virtue of this, they have a moral reason to do so.

One might now object that institutional entities do not have the kind of autonomy that is required for full agency, including moral agency. After all, they are not independent agents.²¹ The issue is whether this fact entails that the actions (or actional events) of institutional entities lack the freedom or autonomy characteristic of the actions of individual persons. There is of course an important controversy as to whether individual persons have the kind of freedom that would be required for it to make sense to hold them morally responsible for their actions. According to the view I take to be the most plausible one, people have the relevant kind of freedom or autonomy just in virtue of being responsive to reasons (Wolf 1990; Fischer and Ravizza 1999). I have already explained how it is that institutional entities respond to reasons, so, on this view, I have already shown how it is that they can have the kind of freedom required for full agency and moral responsibility.

One might think that the fact that institutional entities are not independent agents makes for an additional problem. For, at a fundamental level, an institutional entity has no control over whether individual persons do what would need to be done in order for the institutional entity to comply with its moral duties. So it fundamentally has no control over whether it complies with its moral duties.

This worry does not take seriously enough the point that the actions of individual persons constitute actions of collective entities under relevant conditions. Exercising control is a kind of action. Institutional and other collective entities exercise control in virtue of the actions of persons that constitute their exercising control. For example, a government can threaten with legal sanctions any person who does not do her part in bringing about what the government has decided to do. This example reveals how an institutional entity can exercise control over what people do and thereby exercise control over what it does.

A variety of additional worries could be raised at this point about my thesis that institu­tional actional events qualify as actions. But I hope that the strategy for responding to them is now clear.

Given the considerations I have discussed in this section, I conclude that there is no concep­tual barrier to holding that institutional doers are moral agents and that institutional actional events are actions, properly and strictly speaking. I therefore reject extreme agency individualism. In the next section, however, I consider one last argument for the extreme view. Assuming that we reject the extreme view, I draw the further conclusion that there is no conceptual barrier to holding that institutional entities can have moral obligations and can bear moral responsibility. For if institutional entities are moral agents and can perform actions, then they are relevantly like individual persons in that they are in the scope of obligations. And, like persons, if they fail to act accordingly, they can be responsible for this failing. In the next section, I consider arguments against this further conclusion.

7.7